International Networks and the US-CERN Link

Download Report

Transcript International Networks and the US-CERN Link

Networks and Grids for HEP and Global e-Science
Harvey B. Newman
California Institute of Technology
CANS 2004, Miami
December 1, 2004
Large Hadron Collider (LHC)
CERN, Geneva: 2007 Start
 pp s =14 TeV L=1034 cm-2 s-1
 27 km Tunnel in Switzerland & France
CMS
TOTEM
Atlas
pp, general
purpose; HI
5000+ Physicists
250+ Institutes
60+ Countries
ALICE : HI
LHCb: B-physics
Higgs, SUSY, QG Plasma, CP Violation, … the Unexpected
Challenges of Next Generation
Science in the Information Age
Petabytes of complex data explored and analyzed by
1000s of globally dispersed scientists, in hundreds of teams
 Flagship Applications
 High Energy & Nuclear Physics, AstroPhysics Sky Surveys:
TByte to PByte “block” transfers at 1-10+ Gbps
 Fusion Energy: Time Critical Burst-Data Distribution;
Distributed Plasma Simulations, Visualization, Analysis
 eVLBI: Many real time data streams at 1-10 Gbps
 BioInformatics, Clinical Imaging: GByte images on demand
 Advanced integrated Grid applications rely on reliable,
high performance operation of our LANs and WANs
 Analysis Challenge: Provide results to thousands of scientists.
with rapid turnaround, over networks of varying capability
in different world regions
LHC Data Grid Hierarchy:
Developed at Caltech
CERN/Outside Resource Ratio ~1:2
Tier0/( Tier1)/( Tier2)
~1:1:1
~PByte/sec
~100-1500
MBytes/sec
Online System
Experiment
CERN Center
PBs of Disk;
Tape Robot
Tier 0 +1
Tier 1
10 - 40 Gbps
IN2P3 Center
INFN Center
RAL Center
FNAL Center
~10 Gbps
Tier 2
Tier 3
Tier2 Center
Tier2 Center
Tier2 Center
Tier2 CenterTier2 Center
~1-10 Gbps
Institute Institute
Physics data cache
Workstations
Institute
Institute
1 to 10 Gbps
Tens of Petabytes by 2007-8.
An Exabyte ~5-7 Years later.
Tier 4
Emerging Vision: A Richly Structured, Global Dynamic System
Int’l Networks BW on Major Links
for HENP: US-CERN Example
 Rate of Progress >> Moore’s Law (US-CERN Example)
 9.6 kbps Analog









(1985)
64-256 kbps Digital
(1989 - 1994)
1.5 Mbps Shared
(1990-3; IBM)
2 -4 Mbps
(1996-1998)
12-20 Mbps
(1999-2000)
155-310 Mbps
(2001-2)
622 Mbps
(2002-3)
2.5 Gbps 
(2003-4)
10 Gbps 
(2005)
4x10 Gbps or 40 Gbps (2007-8)
 A factor of ~1M Bandwidth Improvement over
1985-2005 (a factor of ~5k during 1995-2005)
 A prime enabler of major HENP programs
 HENP has become a leading applications driver,
and also a co-developer of global networks
[X 7 – 27]
[X 160]
[X 200-400]
[X 1.2k-2k]
[X 16k – 32k]
[X 65k]
[X 250k]
[X 1M]
[X 4M]
History of Bandwidth Usage – One Large
Network; One Large Research Site
ESnet Accepted Traffic 1/90 – 1/04
Exponential Growth Since ’92;
ESnet Monthly Accepted Traffic 1/90-1/04
Annual Rate Increased from 1.7 to 2.0X
Per Year In the Last 5 Years
300
200
150
100
50
Jul, 03
Oct, 02
Jan, 02
Apr, 01
Jul, 00
Oct, 99
Jan, 99
Apr, 98
Jul, 97
Oct, 96
Jan, 96
Apr, 95
Jul, 94
Oct,93
Jan, 93
Apr, 92
Jul,91
Oct, 90
0
Jan, 90
TByte/Month
250
SLAC Traffic ~400 Mbps; Growth in
Steps (ESNet Limit): ~ 10X/4 Years.
Projected: ~2 Terabits/s by ~2014
Internet 2 Land Speed Records (LSR):
LSR: 6.9 Gbps X 27 kkm 11/08/04
6.6 Gbps
16500km
LSR History – IPv4 single stream
5.6 Gbps
10949km
120
100
4.2 Gbps
16343km
80
5.4 Gbps
7067km
60
2.5 Gbps
0.9 Gbps 10037km
0.4 Gbps 10978km
12272km
40
20
04
Jun-
Nov-
03
Monitoring of the Abilene traffic in LA:
Apr 04
03
Feb-
Oct 03
02
Nov-
0
Apr 02
 Judged on product of transfer speed
and distance end-to-end, using
standard (TCP/IP) protocols, Across
Production Net: e.g. Abilene
 IPv6: 4.0 Gbps Geneva-Phoenix
(SC2003)
 IPv4 with Windows & Linux: 6.6 Gbps
Caltech-CERN (15.7 kkm; “Grand
Tour of Abilene”) June 2004
 Exceeded 100 Petabit-m/sec
 7.48 Gbps X 16 kkm (Linux, 1 Stream)
Achieved in July
 11 Gbps (802.3ad) Over LAN in Sept.
 Concentrate now on reliable
Terabyte-scale file transfers
Note System Issues: CPU, PCI-X
Bus, NIC, I/O Controllers, Drivers
June 2004 Record Network
SC04 BW Challenge: 101.1 Gbps
Petabitmeter (10^15 bit*meter)
Redefining the Role and Limits of TCP
HENP Bandwidth Roadmap
for Major Links (in Gbps)
Year
Production
Experimental
2001
2002
0.155
0.622
0.622-2.5
2.5
2003
2.5
10
DWDM; 1 + 10 GigE
Integration
2005
10
2-4 X 10
 Switch;
 Provisioning
2007
2-4 X 10
1st Gen.  Grids
2009
~10 X 10
or 1-2 X 40
~5 X 40 or
~20 X 10
~Terabit
~10 X 10;
40 Gbps
~5 X 40 or
~20-50 X 10
~25 X 40 or
~100 X 10
2011
2013
~MultiTbps
Remarks
SONET/SDH
SONET/SDH
DWDM; GigE Integ.
40 Gbps 
Switching
2nd Gen  Grids
Terabit Networks
~Fill One Fiber
Continuing Trend: ~1000 Times Bandwidth Growth Per Decade;
Compatible with Other Major Plans (NLR, ESnet, USN; GN2, GLIF)
HENP Lambda Grids:
Fibers for Physics
 Problem: Extract “Small” Data Subsets of 1 to 100 Terabytes
from 1 to 1000 Petabyte Data Stores
 Survivability of the HENP Global Grid System, with
hundreds of such transactions per day (circa 2007)
requires that each transaction be completed in a
relatively short time.
 Example: Take 800 secs to complete the transaction. Then
Transaction Size (TB)
Net Throughput (Gbps)
1
10
10
100
100
1000 (Capacity of
Fiber Today)
 Summary: Providing Switching of 10 Gbps wavelengths
within ~2-4 years; and Terabit Switching within 5-8 years
would enable “Petascale Grids with Terabyte transactions”,
to fully realize the discovery potential of major HENP programs,
as well as other data-intensive research.
Evolving Quantitative Science Requirements for
Networks (DOE High Perf. Network Workshop)
Today
End2End
Throughput
5 years
End2End
Throughput
High Energy
Physics
Climate (Data &
Computation)
SNS
NanoScience
0.5 Gb/s
100 Gb/s
5-10 Years
End2End
Throughput
1000 Gb/s
0.5 Gb/s
160-200 Gb/s
N x 1000 Gb/s
Not yet
started
1 Gb/s
1000 Gb/s + QoS
for Control
Channel
Fusion Energy
0.066 Gb/s
(500 MB/s
burst)
0.013 Gb/s
(1 TByte/week)
0.198 Gb/s
(500MB/
20 sec. burst)
N*N multicast
N x 1000 Gb/s
0.091 Gb/s
(1 TBy/day)
100s of users
Science Areas
Astrophysics
Genomics Data
& Computation
Remarks
High bulk
throughput
High bulk
throughput
Remote
control and
time critical
throughput
Time critical
throughput
Computat’l
steering and
collaborations
1000 Gb/s + QoS
High
for Control
throughput
Channel
and steering
1000 Gb/s
See http://www.doecollaboratory.org/meetings/hpnpw/
Transition beginning now to optical, multiwavelength Community owned or leased
“dark fiber” (10 GbE) networks for R&E
National Lambda Rail (NLR): www.nlr.net
NLR
Coming
Up Now
Initially 4-8 10G
Wavelengths
Northern Route
LA-JAX Now
 Internet2 HOPI
Initiative (w/HEP)
To 40 10G
Waves in Future
Initiatives in: nl, ca,
pl, cz, uk, kr, jp
 + 25 (up from 18) US
States (CA, IL, FL, IN, …)

SC2004: HEP network layout
Joint Caltech,
FNAL, CERN,
SLAC, UF,
SDSC, Brazil,
Korea ….
Ten 10 Gbps
waves to HEP
on show floor
Bandwidth
challenge:
aggregate
throughput
of 101.13 Gbps
achieved
FAST TCP
101 Gigabit Per Second Mark
101 Gbps
Unstable
end-sytems
END of demo
Source: Bandwidth Challenge committee
UltraLight Collaboration:
http://ultralight.caltech.edu
 Caltech, UF, UMich,
SLAC,FNAL, CERN,
FIU, NLR, CENIC,
UCAID, Translight,
UKLight, Netherlight,
UvA, UCLondon, KEK,
Taiwan, KNU (Korea),
UERJ (Rio), USP (Sao
Paolo)
 Cisco
 Next generation Information System, with the network as an integrated,
actively managed subsystem in a global Grid
 Hybrid network infrastructure: packet-switched + dynamic optical paths
 10 GbE across US and the Atlantic: NLR, LHCNet, NetherLight,
UKLight, etc.; Extensions to Korea, Brazil, Taiwan
 End-to-end monitoring; Realtime tracking and optimization;
Dynamic bandwidth provisioning
 Agent-based services spanning all layers of the system
SCIC in 2003-2004
http://cern.ch/icfa-scic
Three 2004 Reports; Presented to ICFA in February
 Main Report: “Networking for HENP” [H. Newman et al.]
 Includes Brief Updates on Monitoring, the Digital Divide
and Advanced Technologies [*]
 A World Network Overview (with 27 Appendices):
Status and Plans for the Next Few Years of National &
Regional Networks, and Optical Network Initiatives
 Monitoring Working Group Report
[L. Cottrell]
 Digital Divide in Russia
[V. Ilyin]
August 2004 Update Reports at the SCIC Web Site:
See http://icfa-scic.web.cern.ch/ICFA-SCIC/documents.htm
 Asia Pacific, Latin America, GLORIAD (US-Ru-Ko-China);
Brazil, Korea, ESNet, etc.
ICFA Report: Networks for HENP
General Conclusions
 Reliable high End-to-end Performance of networked applications such as
Data Grids is required. Achieving this requires:
 A coherent approach to End-to-end monitoring extending to all regions
that allows physicists throughout the world to extract clear information
 Upgrading campus infrastructures.
To support Gbps flows to HEP centers. One reason for under-utilization
of national and Int’l backbones, is the lack of bandwidth to end-user
groups in the campus
 Removing local, last mile, and nat’l and int’l bottlenecks
end-to-end, whether technical or political in origin.
The bandwidths across borders,
the countryside or the city may be much less.
Problem is very widespread in our community, with
examples stretching from the Asia Pacific to Latin America
to the Northeastern U.S. Root causes for this vary, from lack
of local infrastructure to unfavorable pricing policies.
SCIC Main Conclusion for 2003
Setting the Tone for 2004
 The disparity among regions in HENP could increase
even more sharply, as we learn to use advanced networks
effectively, and we develop dynamic Grid systems in the
“most favored” regions
 We must take action, and work to Close the Digital Divide
 To make Physicists from All World Regions Full
Partners in Their Experiments; and in the Process
of Discovery
 This is essential for the health of our global
experimental collaborations, our plans for future
projects, and our field.
PingER: World View from SLAC
S.E. Europe, Russia: Catching up
Latin Am., Mid East, China: Keeping up
India, Africa: Falling Behind
C. Asia, Russia, SE Europe,
L. America, M. East, China:
4-5 yrs behind
India, Africa: 7 yrs behind
Important TCP throughput measured from N. America
to World Regions
for policy
C. Asia (8)
Latin
America
(37)
makers
10000
50% Improvement/year
~ factor of 10 in < 6 years
10000
Edu (141)
1000
1000
Europe(150)
Canada (27)
Mid East (16)
S.E.
Europe (21)
10
100
100
10
Caucasus (8)
Dec-04
Dec-03
1
Dec-02
Dec-01
Africa (30)
Dec-00
Dec-99
India(7)
Dec-98
Dec-97
Dec-96
Jan-96
China (13)
Russia(17)
1
Jan-95
View from
CERN
Confirms
This View
Derived TCP throughput in KBytes/sec
From the PingER project, Aug 2004
PROGRESS in SE Europe (Sk, Pl, Cz, Hu, …)
1660 km of Dark
Fiber CWDM Links,
up to 112 km.
1 to 4 Gbps (GbE)
August 2002:
First NREN in
Europe to establish
Int’l GbE Dark Fiber
Link, to Austria
April 2003 to Czech
Republic.
Planning 10 Gbps
Backbone; dark
fiber link to Poland
CHEPREO Link at SC04:
2.9 (1.95 + 0.98) Gbps
Sao Paulo – Miami – Pittsburgh (Via Abilene)
AmPath
Brazilian HEPGrid: Rio de
Janeiro, Sao Paolo;
Extend Across Latin Am.
HEPGRID and Digital Divide Workshop
UERJ, Rio de Janeiro, Feb. 16-20 2004
NEWS:
Bulletin: ONE TWO
WELCOME BULLETIN
General Information
Registration
Travel Information
Hotel Registration
Participant List
How toTutorials
Get UERJ/Hotel

C++ Accounts
Computer

GridPhone
Technologies
Useful
Numbers
Program

Grid-Enabled
Contact
us:
Analysis
Secretariat

Networks
Chairmen

Collaborative
Theme: Global Collaborations, Grids
and Their Relationship to the Digital
Divide
For the past three years the SCIC has
focused on understanding and seeking the
means of reducing or eliminating the
Digital Divide, and proposed to ICFA that
these issues, as they affect our field of
High Energy Physics, be brought to our
community for discussion. This led to
ICFA’s approval, in July 2003, of the 1st
Digital Divide and HEP Grid Workshop.
More Information:
http://www.lishep.uerj.br
SPONSORS
Systems
CLAF
CNPQ
FAPERJ
UERJ
Sessions &
Tutorials Available
(w/Video) on
the Web
CERNET2 and Key Technologies (J. Wu)
CERNET 2: Next Generation Education and
Research Network in China
CERNET 2 Backbone connecting 20
GigaPOPs at 2.5G-10Gbps
Connecting 200 Universities and 100+
Research Institutes at 1Gbps-10Gbps
Native IPv6 and Lambda Networking
Support/Deployment of:
 E2E performance monitoring
 Middleware and Advanced Applications
 Multicast
Global Ring Network for Advanced Applications Development
www.gloriad.org: US-RUSSIA-CHINA + KOREA Global Optical Ring
 OC3 circuits Moscow-Chicago-
Beijing since January 2004
Aug. 8 2004: P.K. Young,
 Rapid traffic growth with heaviest
Korean IST Advisor to
US use from DOE (FermiLab),
President Announces
NASA, NOAA, NIH and 260+ Univ.  Korea Joining GLORIAD
as a full partner
(UMD, IU, UCB, UNC, UMN…
Many Others)
 Plans for Central Asian extension,
with Kyrgyz Gov’t
> 5TBytes now transferred monthly
via GLORIAD to US, Russia, China
GLORIAD 5-year Proposal (with US NSF) for expansion to 2.5G-10G
Moscow-Amsterdam-Chicago-Pacific-Hong Kong-Busan-Beijing
early 2005; 10G ring around northern hemisphere 2007 (or earlier);
Multi-wavelength hybrid service from ~2008
International ICFA Workshop on HEP
Networking, Grids and Digital Divide
Issues for Global e-Science
Dates: May 23-27, 2005
Venue: Daegu, Korea
Dongchul Son
Center for High Energy Physics
Kyungpook National University
ICFA, Beijing, China
Aug. 2004
Approved by ICFA
August 20, 2004
International ICFA Workshop on HEP Networking, Grids
and Digital Divide Issues for Global e-Science
 Workshop Goals
 Review the current status, progress and barriers to effective
use of major national, continental and transoceanic networks
used by HEP
 Review progress, strengthen opportunities for collaboration,
and explore the means to deal with key issues in Grid
computing and Grid-enabled data analysis, for high energy
physics and other fields of data intensive science, now and in
the future
 Exchange information and ideas, and formulate plans to
develop solutions to specific problems related to the Digital
Divide in various regions, with a focus on Asia Pacific, as well
as Latin America, Russia and Africa
 Continue to advance a broad program of work on reducing or
eliminating the Digital Divide, and ensuring global
collaboration, as related to all of the above aspects.
고에너지물리연구센터
CENTER FOR HIGH ENERGY PHYSICS
Role of Science in the Information
Society; WSIS 2003-2005
 HENP Active in WSIS
CERN RSIS Event
SIS Forum & CERN/Caltech
Online Stand at WSIS I
(> 50 Demos; Geneva 12/03)
 Visitors at WSIS I
Kofi Annan, UN Sec’y General
John H. Marburger,
Science Adviser to US President
Ion Iliescu, President of Romania;
and Dan Nica, Minister of ICT
 Jean-Paul Hubert, Ambassador
of Canada in Switzerland
…
 Planning Underway for
WSIS II: Tunis 2005
Networks and Grids for HENP and
Global Science
 Networks used by HENP and other fields are advancing rapidly
 To the 10 G range and now N X 10G; much faster than Moore’s Law
 New HENP and DOE Roadmaps: a factor ~1000 BW Growth/Decade
 We are learning to use long distance 10 Gbps networks effectively
 2004 Developments: 7+ Gbps TCP flows over 27 kkm; 101 Gbps Record
 Transition to community-operated optical R&E networks (us, ca, nl, pl, cz,
sk, kr, jp …); Emergence of a new generation of “hybrid” optical networks
 We Must Work to Close to Digital Divide
 To Allow Scientists in All World Regions to Take Part in Discoveries
 Removing Regional, Last Mile, Local Bottlenecks and
Compromises in Network Quality are now On the Critical Path
 Important Examples on the Road to Progress in Closing the Digital Divide
 CHINA CNGI Program: CERNET2, CSTNET
 AMPATH, CHEPREO, CLARA and the Brazil HEPGrid in Latin America
 Optical Networking in Central and Southeast Europe
 GLORIAD (US-Russia-China-Korea)
 Leadership and Outreach: HEP Groups in Europe, US, China,
Japan, & Korea
Extra Slides
Follow
ESnet Beyond FY07 (W. Johnston)
AsiaPac
SEA
CERN
Europe
Europe
Japan
Japan
CHI
SNV
NYC
DEN
DC
Japan
ALB
ATL
SDG
MANs
Qwest – ESnet hubs
ELP
NLR – ESnet hubs
High-speed cross connects with Internet2/Abilene
Major DOE Office of Science Sites
Production IP ESnet core
High-impact science core
Lab supplied
Major international
2.5 Gbs
10 Gbs
10Gb/s
30Bg/s
Future phases
40Gb/s
31
CalTech/Newman
FL/Avery
SLAC
Optiputer
Ed Seidel
HOPI
UW/Rsrch Chnl
21 NLR Waves: 9 to SC04
Starlight
NLR-PITT-LOSA-10GE-15
NLR-PITT-LOSA-10GE-14
NLR-SEAT-SAND-10GE-7
SE
A
CHI
PSC
SVL
WDC
CalTech
All lines
10GE
LA
SD
JAX
SC04